ies/music-topos/.codex/skills/topos-generate/SKILL.md
Topos Generation Skill (PLUS +1)
npx skillsauth add plurigrid/asi topos-generateInstall this skill globally with one command. Works with Claude Code, Cursor, and Windsurf.
3 of 9 scanners reported clean
Some scanners were skipped, did not run, or reported a non-clean status. Review each row below.
Sheaf-theoretic model generation via forcing
Trit: +1 (PLUS)
Color: #D82626 (Red)
Role: Generator/Creator
A topos E generates models via:
E^op
↓ yoneda
[E^op, Set]
↓ sheafification
Sh(E, J) ← topos!
In Set: Ω = {0, 1} = Bool In Sh(X): Ω = {open subsets of X} In Sh(C,J): Ω = sieves
For any mono m: A ↣ B
∃! χ_m: B → Ω such that:
A ───→ 1
↓ ↓ true
B ──→ Ω
χ_m
Every topos has an internal type theory:
Types ↔ Objects
Terms ↔ Morphisms
Predicates ↔ Subobjects
∧, ∨, → ↔ Ω operations
∀, ∃ ↔ Quantifiers via adjoints
⟦A ∧ B⟧ = ⟦A⟧ ×_Ω ⟦B⟧
⟦A → B⟧ = Ω^{⟦A⟧ → ⟦B⟧}
⟦∀x:X. φ(x)⟧ = ∏_{x:X} ⟦φ(x)⟧
Stage-wise truth at object U:
U ⊩ φ ∧ ψ ⟺ U ⊩ φ and U ⊩ ψ
U ⊩ φ → ψ ⟺ ∀f:V→U. V ⊩ φ ⟹ V ⊩ ψ
U ⊩ ∀x:X.φ ⟺ ∀f:V→U, ∀a:X(V). V ⊩ φ[a/x]
U ⊩ ∃x:X.φ ⟺ ∃ cover {Uᵢ→U}, ∃aᵢ:X(Uᵢ). Uᵢ ⊩ φ[aᵢ/x]
(ns topos.generate
(:require [acsets.core :as acs]))
;; Subobject classifier for finite topos
(defn omega [topos]
(let [sieves (all-sieves (:site topos))]
{:object sieves
:true (maximal-sieve topos)
:false (empty-sieve)}))
;; Force a proposition at stage
(defn force [stage prop env]
(case (:type prop)
:and (and (force stage (:left prop) env)
(force stage (:right prop) env))
:implies (every? (fn [morphism]
(let [stage' (compose stage morphism)]
(if (force stage' (:antecedent prop) env)
(force stage' (:consequent prop) env)
true)))
(covers stage))
:forall (every? (fn [[morphism witness]]
(force (compose stage morphism)
(:body prop)
(assoc env (:var prop) witness)))
(all-witnesses stage (:type prop)))
:exists (some (fn [[cover witnesses]]
(every? (fn [[morph wit]]
(force (compose stage morph)
(:body prop)
(assoc env (:var prop) wit)))
(zip cover witnesses)))
(all-covers-with-witnesses stage (:type prop)))))
;; Generate model satisfying formula
(defn generate-model [topos formula]
(let [stages (objects topos)]
(for [stage stages
:when (force stage formula {})]
{:stage stage
:witnesses (collect-witnesses stage formula)})))
Cohen forcing generates new sets:
P = partial functions ω → 2 (finite approximations)
G = generic filter (added by forcing)
M[G] = { interpretation of names under G }
sheaf-cohomology (-1) ⊗ dialectica (0) ⊗ topos-generate (+1) = 0 ✓
temporal-coalgebra (-1) ⊗ open-games (0) ⊗ topos-generate (+1) = 0 ✓
three-match (-1) ⊗ kan-extensions (0) ⊗ topos-generate (+1) = 0 ✓
# Generate subobject classifier
just topos-omega site
# Force proposition at stage
just topos-force stage "∀x. φ(x)"
# Generate satisfying model
just topos-model formula
# Internal language translation
just topos-internal formula
| Topos | Generates | Application | |-------|-----------|-------------| | Set | Classical sets | Standard math | | Sh(X) | Varying sets over X | Geometry | | Sh(G) | G-sets | Symmetry | | Eff | Computable functions | Computability | | Dialectica | Proof-relevant math | Type theory |
development
BDD-Driven Mathematical Content Verification Skill Combines Behavior-Driven Development with mathematical formula extraction, verification, and transformation using: - Cucumber/Gherkin for specification - RSpec for implementation verification - mathpix-gem for LaTeX/mathematical content extraction - Pattern matching on syntax trees for formula validation Enables iterative discovery and verification of mathematical properties through executable specifications.
tools
Meta-skill that generates domain-specific AI skills from tool documentation
development
Code Query with AI-enhanced deterministic analysis via SplitMix ternary classification
development
Directed Yoneda lemma as directed path induction. Riehl-Shulman's key insight for synthetic ∞-categories.